Information Technology Reference
In-Depth Information
is analogue of a reliability indicator for models of expert evaluations of
characteristics
{
}
()
(
()
)
l
l
l
L
l
R
μ
x
≡
a
,
a
,
a
,
a
X
= μ
x
;
l
=
1
m
;
, defined over
l
l
1
2
[
]
a
,
.
The analogue of a reliability indicator is connected with fuzziness degree as
follows (with
b
universal set
()
1
): if fuzziness degree of the model is minimum, i.e. is
equal to zero, the analogue of a reliability indicator is maximum and equal to
unity; if fuzziness degree is maximum, i.e. equal to 0.5, the analogue of a
reliability indicator is minimum and equal to 0.25. The certain indicator allows
essential expanding the information obtained on the basis of fuzziness degree
because with identical fuzziness degree the models of expert evaluation of
characteristics can have different values of reliability indicator analogue.
f
x
=
−
x
Example 2.6.
Definition of reliability indicator analogue for models of expert
evaluations of characteristics. Let us consider models of an expert evaluation of
production quality from the example 2.1 and knowledge of students from the
example 2.2. For model of production quality evaluation, analogues of errors of
the first and second kinds are equal to 0.1482, and analogue of a reliability
indicator is equal to 0.7255. For model of evaluation of students' knowledge,
analogues of errors of the first and second kinds are equal to 0.15, and the
analogue of a reliability indicator is equal to 0.7225.
2.8 Ex a mples of Applicati on of Co mplete Ortho gonal Sema ntic Spaces
2.8 Examples of Application of Complete Orthogonal Semantic
Spaces in Problems of Information Analysis and Decision
Making
2.8 Ex a mples of Applicati on of Co mplete Ortho gonal Sema ntic Spaces
Example 2.7.
A multicriterion selection of software. [125—127] The current
software market offers a great many of products for which quality evaluation
systems of characteristics are developed. Complexity of a software selection is
explained by a number of the objective and subjective reasons considered in
details in [128—133].
One of such reasons is use of quantitative and qualitative characteristics, to
measure which the various scales are applied: numerical, ordinal, verbal, etc. At
that, some characteristics can be provided as values of a membership to levels of
linguistic (verbal) scales.
Another reason is discrepancy of quality characteristics, which leads to a selection
ambiguity and makes additive convolution of comparable indicators spurious.
Let us show a solution of software multicriterion selection problem on the basis
of fuzzy conclusion rules.
Let us consider the following characteristics of software quality: modifiability,
studiability, completeness. Let us add software price because it is of interest for a
user along with quality characteristics.
Modifiability
is a characteristic which simplifies introducing of necessary
modifications and updating and includes concepts of expansibility, structuredness
and modularity.
